skill-doctor
Skill1 starsChanged todayMITView on GitHub
Reads unprocessed learnings from project-memory (journal and ledger), triages whether each lesson should update an existing skill, scaffold a new skill via skill-creator, or be skipped, applies anti-bloat edits to the right skill module, audits and optimizes existing skills (diagnosing vague triggers, bloated instructions, orphan assets, edge-case blind spots, and missing verification steps to propose targeted patches), and marks learnings as promoted. Runs periodically inside software-factory-agent or on demand. Use when asked to "skill doctor", "run skill doctor", "update skills from memory", "evolve skills from learnings", "heal skills", "promote learnings to skills", "audit skill health", "audit skill", "optimize skill", or "inspect skill quality".
29 other projects use a version of this.
Adopted by
Most recently taken up by
- liao666brant/skillstheir own version, first seen Nov 15, 2026
- compozy/compozytheir own version, first seen Nov 13, 2026
- isalicema/skill-doctortheir own version, first seen Nov 8, 2026
- qiangjiyi/jiyi-skillstheir own version, first seen Nov 4, 2026
- arthjean/skillstheir own version, first seen Nov 1, 2026
- jongio/skillstheir own version, first seen Nov 1, 2026
Used with
Projects with it also use herdr, diagnosing-bugs and 4 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
ninehillsSkillherdr
Control herdr from inside it. Manage workspaces and tabs, split panes, run commands, read output, and wait for state changes — all via CLI commands that talk to the running herdr instance over a local unix socket. Use when running inside herdr (HERDR_ENV=1).
Also in 3 projects that use this
GoogleSkilldiagnosing-bugs
Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.
Also in 3 projects that use this
NVIDIASkillresearch
Before designing a feature, research how comparable libraries solve the problem, check best practices, known pitfalls, and upstream changes. Use when starting a new feature, investigating an unfamiliar area, or the user asks to research something.
Also in 4 projects that use this
SentrySkillsecurity-review
Finds exploitable application security vulnerabilities in code changes. Use for Warden security scans, appsec review, OWASP-style checks, authentication or authorization bugs, injection, XSS, SSRF, path traversal, secrets, unsafe crypto, webhook verification, open redirects, or sensitive data exposure.
Also in 3 projects that use this
GoogleSkillgrill-me
A relentless interview to sharpen a plan or design.
Also in 4 projects that use this
MicrosoftSkillhandoff
Compact the current conversation into a handoff document for another agent to pick up.
Also in 3 projects that use this
Other versions29 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 29
- nyldn/claude-octopus★ 4.4ksame name
- compozy/compozy★ 2.9ksame name
- dotCMS/core★ 971same name
- warpdotdev/common-skills★ 630same name
- mintuz/skills★ 35same name
- jongio/skills★ 18same name
- evilstar2016/skill-doctor★ 15same name
- Yassimba/loom★ 13same name
- IbrahimSobh/AISkills★ 9same name
- isalicema/skill-doctor★ 7same name
19 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository38 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-orchestratorSkill
- algorithmic-complexity-analyzerSkill
- api-securitySkill
- architectural-planningSkill
- bdd-gherkin-testerSkill
- caching-queues-reviewerSkill
- code-reviewerSkill
- csharp-developerSkill
- design-pattern-detectorSkill
- detect-overengineeringSkill
- github-security-scanner-hookSkill
- java-developerSkill
- legacy-code-modernizationSkill
- llm-as-judgeSkill
- load-memorySkill
- memory-analyzerSkill
- nodejs-executorSkill
- playwright-testerSkill
- powershell-writerSkill
- python-api-writerSkill
- python-developerSkill
- python-linterSkill
- rag-pipelineSkill
- react-developerSkill
- remote-pod-log-analyzerSkill
- researcher-poc-testerSkill
- save-memorySkill
- scan-halucinated-testsSkill
- scan-repo-readmeSkill
- skill-creatorSkill
- sql-connectorSkill
- sql-executorSkill
- sync-github-forksSkill
- test-data-creatorSkill
- trace-code-contextSkill
- troubleshoot-and-fixSkill
- typescript-developerSkill
The file
We split this skill into the rules it gives an agent
12 rules. 0 are common, 3 are distinctive, and 9 are very specific to this project.
0
Common
Most of the 25 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 17
After debugging a tricky platform, security, framework, or tooling trap that future sessions must not repeat.
- Very specific: line 18
When auditing or optimizing an existing skill's SKILL.md, references/, and scripts/ for vague triggers, instruction bloat, edge-case blind …
- Very specific: line 19
When asked to "run skill doctor", "update skills from memory", "evolve skills from learnings", "promote learnings to skills", "audit skill …
- Very specific: line 35
If 0 pending learnings exist, run audit-skills to verify repository health and exit cleanly.
- Very specific: line 55
Add new pitfalls, patterns, or guardrails to the matching topical submodule (core/, runtime-and-memory/, concurrency/, security/, testing/, …
- Very specific: line 57
Update the relevant section (## Workflow, ## Common Pitfalls, or references/*.md), or harden helper scripts in scripts/.
- Very specific: line 59
Never embed local user directory paths (C:\Users\..., /home/...) or secrets in skill docs or scripts.
- Very specific: line 67
When CREATE_NEW_SKILL is selected:
- Distinctive: line 82
Run the skill health audit:
- Very specific: line 90
If any Python/JS/TS scripts were created or modified, verify low over-engineering risk (score < 40):
- Distinctive: line 100
Use this mode when asked to audit, diagnose, or optimize an existing skill (or all skills):
- Very specific: line 108
Read the target skill's SKILL.md, references/*.md, and scripts/*, evaluating against references/skill-evolution-rubric.md:
“Common” is measured: kept by at least 6 in 10 of the 25 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Skill Doctor
Closed-loop continuous improvement skill that converts hard-won project-memory learnings into durable repository skills—updating existing skill modules, creating new skills when recurring capability gaps emerge, and auditing all skill health.
When to Use This Skill
- Periodically at the end of a
software-factory-agentrun (when>= 3unprocessed learnings accumulate or Phase 8 remediation was triggered). - After debugging a tricky platform, security, framework, or tooling trap that future sessions must not repeat.
- When auditing or optimizing an existing skill's
SKILL.md,references/, andscripts/for vague triggers, instruction bloat, edge-case blind spots, or missing verification steps. - When asked to
"run skill doctor","update skills from memory","evolve skills from learnings","promote learnings to skills","audit skill health", or"optimize skill <name>".
Mode A — Memory-Driven Skill Evolution
Step 1 — Harvest Unprocessed Learnings
Retrieve pending journal and ledger learnings via the project-memory MCP tool get_unprocessed_learnings, or run the diagnostic CLI:
.\.venv\Scripts\python.exe .github/skills/skill-doctor/scripts/skill-doctor.py diagnose(On POSIX, substitute .\.venv\Scripts\python.exe with .venv/bin/python.)
If 0 pending learnings exist, run audit-skills to verify repository health and exit cleanly.
Step 2 — Triage & Route Each Learning
Consult references/skill-evolution-rubric.md and classify each pending memory into one of three actions:
| Classification | Criteria | Action |
UPDATE_EXISTING_SKILL | Matches an existing skill domain and adds a reusable pitfall, guardrail, command convention, or pattern not yet documented. | Edit the target skill module (or SKILL.md / references/) and mark promoted. |
CREATE_NEW_SKILL | >= 2 learnings cluster around a repeatable technology, workflow, or domain not covered by any existing skill. | Scaffold via skill-creator, populate modules, register in factory/orchestrator/README, and mark promoted. |
SKIP_TRANSIENT | Already documented in the target skill (ALREADY_COVERED) or one-off task/ticket state (SKIP_TRANSIENT). | Do not bloat skills; mark skipped so it is not re-triaged. |
Step 3 — Apply Skill Updates Safely (Anti-Bloat Guardrails)
- Modular Knowledge-Pack Skills (e.g.,
java-developer,python-developer,typescript-developer,react-developer,csharp-developer,powershell-writer):- NEVER bloat the lean
SKILL.mdrouter (keep under 60 lines). - Add new pitfalls, patterns, or guardrails to the matching topical submodule (
core/,runtime-and-memory/,concurrency/,security/,testing/, etc.) orreferences/.
- NEVER bloat the lean
- Task / Reference Skills:
- Update the relevant section (
## Workflow,## Common Pitfalls, orreferences/*.md), or harden helper scripts inscripts/.
- Update the relevant section (
- Zero Personal Paths & Secret Hygiene:
- Never embed local user directory paths (
C:\Users\...,/home/...) or secrets in skill docs or scripts.
- Never embed local user directory paths (
- Bump Semantic Version:
- Increment
versionin the updatedSKILL.mdfrontmatter (PATCHfor pitfalls/clarifications;MINORfor new sections or scripts).
- Increment
Step 4 — Scaffold New Skills (When Warranted)
When CREATE_NEW_SKILL is selected:
- Run
skill-creatorscaffolding:node .github/skills/skill-creator/scripts/scaffold-skill.js <skill-name> .github/skills [--modular] - Populate
SKILL.md(with 5–8 realistic trigger phrases indescription) and supporting modules/references/scripts using the harvested learnings. - Register the new skill in:
.github/agents/software-factory-agent.md(skill count + routing matrix).github/skills/agent-orchestrator/scripts/orchestrator-context-manager.py(stage mapping if applicable)README.md(Skills Catalog table, category section, and directory tree)
Step 5 — Validate & Mark Learnings Promoted
- Run the skill health audit:
.\.venv\Scripts\python.exe .github/skills/skill-doctor/scripts/skill-doctor.py audit-skills - Validate all
SKILL.mdfrontmatter and YAML files:npm run validate:skills --prefix mcp/sql-connector - If any Python/JS/TS scripts were created or modified, verify low over-engineering risk (
score < 40):.\.venv\Scripts\python.exe .github/skills/agent-orchestrator/scripts/orchestrator-context-manager.py check-overengineering --file <modified-script> - Mark processed memory IDs via
project-memoryMCPmark_learning_promoted(or CLIskill-doctor.py mark --ids <id1,id2> --status promoted --target-skill <skill>).
Mode B — Skill Auditor & Optimizer
Use this mode when asked to audit, diagnose, or optimize an existing skill (or all skills):
- Run Static Quality & Structural Audit:
(Omit.\.venv\Scripts\python.exe .github/skills/skill-doctor/scripts/skill-doctor.py audit-skills --skill <skill-name>--skill <skill-name>to audit the entire.github/skills/catalog; add--strictto fail on warnings.) - Inspect Skill Files & Diagnose Semantic Weaknesses:
Read the target skill's
SKILL.md,references/*.md, andscripts/*, evaluating against references/skill-evolution-rubric.md:- Vague Trigger Descriptions: Missing concrete
"Use when..."scenarios or 5–8 quoted trigger phrases. - Bloated / Noisy Instructions: Modular
SKILL.mdrouters exceeding ~65 lines, redundant prose, or unlinked orphan files inreferences//scripts/. - Edge-Case Blind Spots: Bare
python/pycalls instead of.venv, unsafe shell quoting, or missing platform caveats (Windows vs POSIX). - Missing Verification Steps: Task skills with
scripts/lacking expected output, exit-code, or post-edit validation commands (and runcheck-overengineeringon scripts).
- Vague Trigger Descriptions: Missing concrete
- Propose & Apply Targeted Patches:
- Present or apply minimal, localized diffs (do not rewrite working sections from scratch), bump
versioninSKILL.mdfrontmatter, and re-runaudit-skills --skill <skill-name>+npm run validate:skills --prefix mcp/sql-connectorto verify zero errors.
- Present or apply minimal, localized diffs (do not rewrite working sections from scratch), bump
Additional Resources
- Skill Evolution Rubric — Triage decision matrix, deduplication rules, modular router protection, and versioning guide.
- Skill Doctor CLI — Automated diagnostic, schema migration, skill audit, and promotion-marking script.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

